Introduction

Literacy, numeracy, and life skills are basic skills needed to engage individuals and harness their inputs towards national development, hence the varied responses from global and national actors, especially in insurgency-ravaged areas such as north-east Nigeria. The Nigerian northeast region has been embroiled in the insurgency since 2002 when the Boko Haram sect, which literarily translates that Western education is forbidden, took up arms against the state and education. The modus operandi of driving their ideology has left in its wake massive destructions of infrastructure and human displacements. Among areas closed are learning spaces, which have been argued to threaten the progress of children, youth, and adults, particularly those from more disadvantaged families (Agarwala et al. 2022). These identified actors are deeply committed to rebuilding the northeast region plagued by insurgency and counter-insurgency activities, which have led to school closures, destruction of social infrastructure, massive displacements and disruptions, and the like (Jacob and Ensign 2020).

It is also conceivable to look beyond the effects of insurgency as causes of poor literacy, numeracy, and life skills. This is because teaching methods, pedagogical designs, and motivations in reaching out to stakeholders in their scattered, dispersed, and uncoordinated domains could be more serious sources of concern. And the insurgency period, coupled with the clawing roles of COVID-19, had cultured grounds for these to thrive. There is, thus, the need to reengage appropriate pedagogy and technologies as such situations are argued to open new pedagogical avenues for meeting basic learning needs (Azevedo et al. 2020; Karam 2020; Maity et al. 2021). COVID-19 and insurgency-related situations have given more credence to this than ever before since learning areas are inaccessible. Learning guide experts prefer the comfort of their homes, the flexibility and depth of information left at the disposal of the learner courtesy of an induced-tech-pedagogy (Ferguson et al. 2020; Rasmitadila et al. 2020) as well as the hard-to-reach segment who are cut out because of their socioeconomic cum political narratives (Ho and Laflin 2009) to risking the unsecured terrains. It should be emphasized here that technology should be seen beyond its digital frame to its sociocultural and political appendages in shaping reality (Topping et al. 2022). The COVID-19 pandemic, together with mobility and global security concerns, has brought about a transformation in the socio-educational, cultural, and psychosocial approaches of institutions and individuals. Despite the challenges, this has presented an opportunity to reassess and improve these areas. The changes brought about by these factors have created a platform for positive growth and development (Obukoadata et al. 2021; Maity et al. 2022). The pendulum has shifted in favor of digitalization and interaction, and adjustments are needed for stakeholders to continue to function optimally. In short, one takeaway, which is difficult to classify as either positive or negative is that of ‘isolatism’ in the knowledge conundrum.

Against this backdrop, the SENSE-TRI model was conceived to mitigate the pandemic and insurgency. Transactional radio instruction (TRI) suggestively addresses these yearnings through calls for the effective deployment of synchronous radio instructional broadcast (Adebanjo 2018), simultaneity in multitasking the multisensory capabilities of learners and giving them room to probe beyond just the nuances of classroom engagements (Johnson and Aragon 2003; Harris et al. 2010; Kristanto 2016) when the educational system is met with challenges of insurgency and pandemic such as the COVID-19. Deploying the same designs in Nigeria would amplify arguments on the deployment of radio as a potent means of development communication in Africa and its use in mobilizing the citizenry in consonance with the characteristics features of the radio medium (Cazalet 2017).

While the SENSE-TRI is not completely novel, its application in this study involves the aggregations of several other models and approaches as one. Its deployment to a two-pronged challenging situation – insurgency and pandemic – makes the study a compelling one. Other studies had focused on deploying or replication one of such interventions, but this had gone on to deploy it to two challenging quotients. However, the question is, “Does participation in the TRI intervention implemented in north-east Nigeria lead to improved literacy outcomes compared to learners not involved in the SENSE-TRI program?”

This paper aims to investigate whether learners engaged in the TRI intervention exhibit superior literacy outcomes compared to their non-SENSE-TRI counterparts. The underlying assumption is that by building upon the proven effectiveness of TRI in previous studies, modifications to its core principles would likely yield even more pronounced success.

By doing so, this paper adds to the body of knowledge on the deployment of TRI, particularly in challenging settings such as those prevalent in northeast Nigeria (Agarwala et al. 2022). The remainder of this introduction provides a theoretical justification for the TRI approach and discusses the empirical evidence available on this practice from around the world. We then describe the study setting and the research design. Subsequently, we present and discuss our empirical results and conclusions.

The theoretical rationale for SENSE-TRI programming

The Strengthening Education in Northeast States (SENSE) Project, funded by USAID, as designed, addresses daunting challenges facing the basic education system in Adamawa and Gombe states. It was a three-year basic education project whose goal was to improve identifiable outcomes for learners in lower primary grades through teacher training, improved community engagement, and the provision of teaching and learning materials (Martens et al. 2020). The Project partners with various state, national, and international institutions to analyze and develop strategies for a range of scenarios. Utilizing an approach tailored to each situation, the Project addresses any challenges stemming from a lack of uniformity. Continuously evaluating effectiveness, identifying areas for improvement, and scaling successful initiatives for wider adoption, the Project remains poised to replicate successful outcomes on a national scale.

The project scope was expanded force majeure in response to the COVID-19-induced school shutdown in late March 2020 to include the radio element, thus subsuming the transactional interactive radio instruction (TRI) component to reach learners in the safe confines of their households with the catch-phrase “schooling on air” or “homeschooling”. Several educational contents meant for normal classroom engagements were converted to suit the TRI pedagogy. These include thirteen weeks of “Mu Karanta” and “RANA” aired to learners in the two states from August to October 2020.

The Transactional Radio Instruction (TRI) program seeks to get learners to connect with the radio lesson by designing the lessons as radio-drama series accompanied by easy-to-memorize songs and storytelling, and synchronized workbook activities that induce engagement, excitement, and entertainment. TRI is made up of four core components (or 4 S) which are sound pedagogy, stories, songs, and synchronized workbooks. The learning materials were designed along the early grade reading (EGR) best practice model relying heavily on the read-aloud approach, that is, interspersed with songs. Each episode is aired twice a week. This method promotes learner involvement, increases anticipation, and helps the learners to remember the story (message) with the aid of short written activities such as letter identification, tracing, completing, and copying. This approach is synchronized with the outcomes of teachers’ training and improved community engagements.

The TRI programming aligns with the dictates of the UNICEF Assessment, Communication Analysis, Design and Action (ACADA) planning model developed in the eighties and nineties and has roots in the early effective implementation of the National Polio Immunization (NPI) programs by transnational governments. It focuses on the Triple-A planning cycle of assessment, analysis, and action (Anaeto et al. 2008). Assessment of the program under this model involved ‘documentation of the status of awareness or intervention, evaluation of SWOTS, lessons, issues, problems, participants, behaviors, credible channels and so on’ so as ‘to identify what information is missing and to design and carry out research to fill in the gaps’. For communication analysis of the program, the emphasis is placed on problem analysis, formulation, determination, and the conduct of behavior, participation, channel, and media analysis as well as identification of indicators to drive the campaign and development of monitoring and evaluation indicators. While ‘action’ is the end point of the intervention and requires an evaluation to see if set goals are met.

The completion of the SENSE-TRI implementation up to the baseline data analysis and intervention stage is a significant achievement. However, there is still a need for further analysis to assess the impact of deploying TRI in knowledge transactions, particularly in areas where schools are unavailable due to insurgency or pandemics. The paper benefits from the support of the ACADA planning model, which provides a strong theoretical framework for the study.

Experiences with TRI and technologies from around the World

The argument has been canvassed about the deployment of appropriate technologies during a pandemic such as COVID-19, and the pace with which several countries have embraced such digital learning pedagogy. Inherently, it was subsumed that despite this super role importance, virtual teaching cannot be replaced by physical contact, but may only play supplementary roles, and establish a strong relationship between teachers and students (Maity et al. 2021). This helps, but accentuates earlier studies (Agarwala et al. 2022) that deploying technologies under the TRI approaches can eliminate the possibilities of religious, cultural, social, and economic bias between the students and the teachers since such technologies are impersonal. Thus, when identified threats are immediately addressed, technologies channeled through TRI inadvertently improve students’ overall educational performance by reflecting paradigm shifts in educational models, cultural heritages, experiences, and perspectives. (Gay 2002; Goldman 2018; Maity et al. 2021).

TRI-induced tech-modeled learning is argued to help strengthen developmental ideologies and the effectual participation of stakeholders in places where there are no schools (Jacob and Ensign 2020). However, we cannot discountenance the importance of the blended approach since no study has established the potential of only TRI in solely achieving learning goals to the exclusion of other pedagogies. Our study focused on the premise that the integration of SENSE ideologies serves as a valuable supplement to the TRI model. Furthermore, our observations indicate that the implementation of blended learning, as proposed by the (Geta and Olango 2016) study, can greatly improve students’ writing skills in courses. As a result, the adoption of a blended learning approach to reinforce the benefits of the TRI model in a highly effective manner is further canvassed. Technology would always be a decimator for either the blended or TRI approaches resulting in displacement as presumed by the technological determinism discourse of Marshall McLuhan, while users may become too media inclined as argued by the media equation theorists within the convulsed cybernetic tradition milieu (Griffin 2012).

We have also observed in line with Maity et al. (2022) that opportunity and capacity to adapt these digital approaches is one thing, and adaptability to the new paradigm and level of virtual engagements is another, especially as it has to do with the new mode of tech-driven teaching and learning in the face of insurgency and pandemic. Adaptability could be influenced by internet penetration, issues of electricity, mobile network, internet glitches, and level of available income to spare for such endeavors, especially in countries where there are no definite policies in this regard as well as the willingness of the key players to get in tune with the new approaches. For success to be recorded, there must be a conscious blend between opportunity and capacity on one hand and adaptability on the other.

Earlier findings suggest most of the radio technology-driven programs have well-defined educational objectives, production, and presentational procedures consistent with best practices. They also would normally fall short in terms of capabilities for interaction among listeners, moderators, and guests (Agarwala et al. 2022), and this impinges on cognition, attitude formation, and behavioral modifications (Vanderplank 2010; Patrick 2018). Nonetheless, intensity in the use of technologies would enhance the potential across media ecologies (Haddad and Draxler 2002).

The implications for technological uses are also pulsated on several levels. These include the need for teachers to know that online and blended learning may be substituted for regular instruction and as such they should be empowered to become digitally inclined in all situations as the need arises. Empowerment could come in the form of exposing them to “computer-supported collaborative learning, educational games, and computer-assisted instruction delivered into homes” (Topping et al. 2022). Suggesting that a blended approach still hold some more advantage, especially in developing countries.

In the northeast of Nigeria, the SENSE-TRI program proposed several benchmarks, set goals, and evaluated, and reevaluation the parameters to accommodate new challenges and legwork. This process of appraising the extent of work directed at deepening literacy and numeracy skills among certain categories of persons is expected to benefit from the deployment of radio technologies. While the rampaging COVID-19 is seen as a key decimator in this study and to all things that had been deemed normal before the moment, teachers, who are the major stakeholders, must align with instructional strategies, challenges, support, and motivation (Karam 2020; Kirshner 2020; Rasmitadila et al. 2020) as well as the readiness of technology, wound around the humanist curriculum in delivery.

Pedagogical contents are driven by the butt of technology. Also, the effective integration of available technologies into the pedagogical ecology of the people is considered more sacrosanct and impactful since it deals with the nuances of the people (Harris et al. 2010; Obukoadata et al. 2020). The integration of radio technology which has been argued to be immensely powerful among rural dwellers, is also quite instructive. Such integration is a product of the teachers’ concurrent knowledge and the goal of the curriculum content, expressive in meeting the learning needs and educational goals of the stakeholders. Similarly, a designed media literacy education curriculum is expected to focus on specific media analysis skills of succinctly identifying the main ideas, purpose, and structural features of a broadcast. Thus, an enriched integration of media literacy skills into the curriculum is likely to yield optimal results (Hobbs and Frost, 1998; Ilboudo et al. 2018; Hodges et al. 2020). This investigation is based on the notion that technological methods are not fundamentally novel but rather an evolution of the various educational aspects of traditional in-person instruction. It is posited that these methods provide increased versatility in completing tasks, foster independence in learners, and promote greater self-control as pupils can adhere to their motivational inclinations (Topping et al. 2022).

Instances are evident in the 1993 Bolivian experiment, where, instead of just focusing on the young caregivers, parents, child development specialists, and health educators in the learning cycle, more funds were deployed towards radio-based instructional programs (Bosch and Crespo, 1995); the development of geography knowledge of students (Caldis 2018) as well as delivering educational services to youths and children in crisis environments (Driscoll 2010; Carlson and JBS International 2013; Education Development Center 2019).

Globally, radio has been argued to be very definitive in mobilizing for change. A Nicaraguan experience in the early 80 s proved this assertion when deployed to helping the students understand and appreciate mathematics through the use of textbooks and radio treatments. This had significant positive effects on the achievements of the students even though it is not suggestive enough for either textbooks or radio instructions to be solely sufficiently powerful to induce this result (Jamison et al. 1981; Johnson and Aragon 2003). One stronghold of the radio tool in providing learning engagement is that it is the voice of the poor and can help boost development (Madamombe 2005), enrich the learning styles of the recipients who are immersed in its use for other things other than learning and thus strike a new fancy for them (McGovern et al. 2017). Radio also remains a powerfully cheap medium of communication over distant places because of its ubiquitous nature in crisis periods globally (Uduma and Obukoadata 2016; Geta and Olango 2016; Okeke et al. 2020; Ullah 2020). Despite these strong points, the focus should not be to institutionalize interactive radio, but to institutionalize interactivity which helps provide a marked departure from the classroom environment (Olsson 1994). This is necessary, as, despite the significant potential of radio as an educational broadcasting tool, awareness and utilization for the same purpose by a segment of society is still low. There is, therefore, the need to institutionalize the interactivity of the interactive radio (Olumorin et al. 2018) as well as negotiate effective teaching and learning via a classroom away from the classrooms, and pedagogy away from pedagogies (Crabtree and Sapp 2004).

TRI has been established to be effective across cultures and regions of several European cities (Elliott and Lashley 2017), what about the Nigerian north-east? Although some regions would present diverse challenges, the expected goals are nonetheless the same across processes (Kozma, 1991; Scott-Kassner, 1999; Gass 2002). Cultural deployment to the SENSE-TRI program is, therefore, not without distraction from this, and therefore, necessary to ensure that the SENSE-TRI is still relevant in meeting set goals and visions.

Methodology and materials

Study region

The SENSE-TRI study provided a valuable opportunity to evaluate the educational advancement of primary three students in Adamawa and Gombe states. Despite being situated in an area affected by insurgency, these states were selected for their relative stability and ease of access for research purposes. Additionally, the two states naturally provided a significant study area as the out-of-school children were among the highest in the country. See the location in Fig. 1.

Fig. 1
figure 1

Map of Nigeria showing the TRI study areas.

Full size image

Sampling method and participants

A quasi-experimental design was used to select schools and students for the research. The target population consisted of all schools participating in the SENSE intervention program, which aimed to improve educational outcomes. The first step in the sampling process was the random selection of schools from this population to form the treatment group. The treatment group included schools and students who were exposed to the SENSE-TRI program.

To create a comparison group, schools and students were selected from the general population of schools that did not have exposure to the SENSE-TRI program. The selection process ensured that the comparison group schools were similar to the treatment group schools. Specifically, comparable schools located in close proximity to the treatment group schools were chosen. This proximity criterion aimed to minimize potential confounding variables and increase the comparability of the two groups. Figure 1 depicts this selection process.

To ensure the generalizability of the results in both states, the sample size was determined using a 95% confidence interval and a 5% margin of error. This calculation indicated a minimum requirement of 381 students. However, for ease of division into two groups, the sample size was approximated to 400 students, with 200 students assigned to the treatment group and 200 students to the comparison group.

In terms of participant selection, a quota-random-walk sampling method was employed to choose grade 3 (primary 3) students from all households located near the SENSE intervention schools. This sampling method ensured that students from different households in the vicinity had an equal chance of being selected. The selected students were then divided into the treatment group and the comparison group.

Data was collected from a total of 200 primary grade 3 learners in Gombe and 200 learners in Adamawa, two states included in the study. In each state, the number of participants was equally divided between learners in schools that belonged to the treatment group and learners from comparable non-program schools. It is worth noting that due to the pandemic, face-to-face teaching did not take place in any of these schools. Therefore, all participants were formally enrolled in school but practically learning from home.

Instruments and data collection

After the radio broadcasts ended in mid-December 2020, an abridged version of the Early Grade Reading Assessment (EGRA) was conducted. In line with the study objective of determining whether and how the TRI model would enhance knowledge acquisition during insurgency and pandemic where there are no schools. The EGRA consists of five subtasks. These include: letter-sound identification (identification of 29 letter sounds), syllable-sound identification (identification of 21 syllable sounds), familiar word reading (decoding of 18 words familiar to the learners), invented word reading (decoding of 12 invented words), and reading comprehension (answering 4 questions about the connected text).

As the tasks were not timed, scores were not reported as correct answers per minute. Instead, the scores were based on the number of correctly answered items. The EGRA was administered to the target learners, specifically grade 3 learners, by a consistent group of trained enumerators. These enumerators followed a random-walk approach, moving from one location to another, to sample learners across different areas devoid of bias. Nonetheless, variability may likely exist for enumerators. These variabilities could be psychosocial, sociological, and cultural, and have the propensity to impact their activities. Data collection was facilitated using Android tablets preloaded with the abbreviated version of the EGRA on the Tangerine platform, which served as the tool to administer the assessment.

Data analysis and results

The TRI program was evaluated by comparing the scores (=correctly answered items) of learners in the treatment group with the scores of learners from a comparison group that did not participate in the program. This was done, first, by comparing EGRA results between the treatment and comparison group using a nonparametric equality-of-medians test, given the high percentage of students scoring at the extremes of the grading scale. We then reconsider these comparisons between groups by controlling for student background characteristics. To account for the lower and upper bounds of the test scores, this is done using Tobit regression models as given below:

$$y_i^ \ast = x_i^\prime \,\beta + \varepsilon _i$$
(1)
$$y_i = \left\{ {\begin{array}{l} {a\,if\,y_i^ \ast \le a} \\ {y_i^ \ast \,if\,a\, <\, y_i^ \ast\, <\, b} \\ {b\,if\,y_i^ \ast \ge b} \end{array}} \right.$$
(2)

Here, \({{{\boldsymbol{y}}}}_{{{\boldsymbol{i}}}}^ \ast\) is the unobserved, latent outcome variable, but we can only observe yi, which is truncated and has a as is the lower bound (a) and an upper bound (b). \({{{\boldsymbol{x}}}}_{{{\boldsymbol{i}}}}^\prime\) is a vector of explanatory variables, β is a vector of unknown parameters, while the εi is a disturbance term. The latent outcome variable \({{{\boldsymbol{y}}}}_{{{\boldsymbol{i}}}}^ \ast\) rests on the classical linear model assumptions of a normal, homoscedastic distribution with a linear conditional mean (Wooldridge 2000). Our regression models include dummies for Local Government Areas (LGAs) to account for regional differences that might be responsible for considerable variation in individual students’ scores as well as the enumerator’s variability and biases. Results are reported based on robust standard errors. The parameters of the model were estimated using the maximum likelihood method and can be implemented in Stata or R software.

Student characteristics

Table 1 summarizes the demographic characteristics of the learners in the comparison and treatment groups. It also includes the p-value of the test for differences between treatment and comparison groups. As it turns out, the groups are statistically similar in many ways. They do, however, differ in some important observed characteristics. For instance, students in the treatment group are, on average, older than students in the comparison group; they are more likely to speak Hausa at home; and tend to have more books at home, indicating a family background more conducive to study and be educated. All of these characteristics can generally be expected to correlate positively with Hausa literacy skills; but would they correlate easily with English literacy skills? In both groups, there were more girls than boys, which is very significant to the overall results, considering the nuances of the region where the females are less likely to be exposed to Western education especially as the menace of Boko Haram insurgency continues to grow.

Table 1 Demographic characteristics of the study (comparison vs treatment).
Full size table

Differences in reading skills between study groups

Table 2 gives an overview of the subtask scores those students achieved, for the pooled sample and the treatment and comparison group separately. The table also contains the observed minimum and maximum scores, which correspond to the theoretical minimum and maximum scores for the assessment tool. It is clear from the table that students in the treatment group vastly outperform students in the comparison group. This is true for all subtasks and all reported indicators used in the study. There are students scoring zeros in the percentile distribution, mean scores, and median scores. For instance, in the basic subtask letter sound identification, 26% of learners in the comparison group scored zero, while only 3% of students in the treatment group did so. The average observed subtask mean score in the former group is 8.76, while it is 24.01 for the treatment group. Finally, the subtask median score is 4 for the comparison group and 26 for the treatment group. Similar patterns can be observed for all subtasks. Figures 2 and 3 summarize the distribution of subtask scores graphically through box plots and histograms, respectively. Both graphs clearly show the differences in the distribution of scores between the two study groups, with learners from the treatment group clustering at the high end of the scale, and learners from the comparison group clustering at the low end of the scale. Unsurprisingly, a Pearson Chi-square test for differences between the two study groups turns out to be highly statistically significant for all subtasks (Pearson Chi-square test statistics reported at the bottom of Table 2).

Table 2 Differences in reading skills between study groups.
Full size table
Fig. 2
figure 2

Box plots: distribution of subtask scores, by study group.

Full size image
Fig. 3
figure 3

Histograms: distribution of subtask scores, for the comparison group (top row) and treatment group (bottom row).

Full size image

Differences in reading skills between groups, controlling for student characteristics

To account for the observed differences in student characteristics (see Table 1), we reconsider the differences between the two study groups after including a set of control variables. We use a Tobit regression model for this purpose because test scores are clustered at the upper and lower end of the possible range of scores. The results are presented in Table 3. Even after controlling for student characteristics, the difference in the test scores between treatment and comparison group remains large and statistically significant for all subtasks. The estimation coefficient for the treatment group is estimated to be 17.944 and statistically significant at the 5% level for letter-sound identification; the respective estimation coefficients for syllable sounds, familiar word reading, invented word reading, and reading comprehension are all significant at the 1% level and amount to 18.123, 16.695, 17.311, and 4.041, in that order. The results furthermore show higher scores for students who speak Hausa at home, and lower scores for students who have suffered hunger in the past week (other things being equal) preceding the collection of data because of the endemic poverty and perception about getting formal education in the northeast of Nigeria. Notably, the correlation between the treatment group and speaking Hausa at home is negative, important in size, and statistically significant. Hence, while students speaking Hausa at home have generally higher scores than those not speaking Hausa at home, the difference is smaller (or even statistically non-existent) for students in the treatment group. The marginal predictions of these differences are depicted in Fig. 4, together with their respective 95% confidence intervals. The results of the reading comprehension subtask are particularly noteworthy. In the comparison group, students who speak Hausa at home perform significantly better than those who don’t. However, in the treatment group, the opposite is observed, with students who don’t speak Hausa at home performing better than those who do.

Table 3 Results of the Tobit regression model, based on study group.
Full size table
Fig. 4
figure 4

Predicted subtask scores: predicted margins with 95% Confidence Intervals.

Full size image

Model fit and regression diagnostics

The Pseudo- R2 value at the bottom of Table 3 indicates that the model could explain only between 11.7% and 16.5% of the observed variance. This is not surprising in the education domain. However, the link test for model specification showed no evidence of model misspecifications, except in the case of the model for letter sound scores, where the transformed independent variable has a significant p-value of 0.025. The results of this test are not included in the table.

Alternative specifications and robustness

The number of available books at home was found not to be statistically correlated with the test scores of the children. As the variable contains a relatively high percentage of missing values, it was therefore removed from the final model. To check the robustness of the results of the final model, we varied the definition of our main outcome of interest, that is, we varied the frames of what improved literacy would imply within a randomized context. Instead of using a dummy variable, assigning a learner to either the treatment or comparison group, we use the self-reported frequency where a learner listened to the TRI radio broadcast. This enhances active program participation instead of mere participation. We consider allowing the respondents listen to the broadcast at least weekly to be an indicator of sufficient program exposure. The results are presented in Table 4 and indicate a statistically significant correlation between exposure to program content, program participation, and test scores via interactions with state of residence, age, socioeconomic decimators, and speaking Hausa at home. The main effect is only significantly noticeable for the subtask invented word reading. To evaluate the effects of listening to the radio broadcast, all variables that are part of the interaction were considered. We, therefore, in Fig. 5, present the estimated marginal predictions for learners listening or not listening to the broadcast at least every week (all other variables are evaluated as observed). The results confirmed that learners who listen to the broadcast every week outperform those who do not listen to the broadcast at such a consistent frequency.

Table 4 Results of the Tobit regression model, based on self-reported radio listening.
Full size table
Fig. 5: Predicted subtask scores, alternative model: predicted margins with 95% Confidence Intervals.
figure 5

“Comp” refers to learners not listening to the radio broadcast at least weekly, and “Treat” to learners listening to the radio broadcast at least weekly.

Full size image

Discussion

Overall, our results show that learners who participated in the SENSE-TRI intervention being rolled out in north-east Nigeria, have better literacy outcomes than other non-participant learners. This is the case for all tested skill sets (letter sound identification, syllable sound identification, familiar word reading, invented word reading, and reading comprehension), and holds up even after including the control variables.

While these results are encouraging, the absence of baseline data and the lack of random assignment of learners to the two study groups make it impossible to wholly attribute the better learning outcomes to the treatment group. In particular, an important question that remains is whether the socio-demographics, sociological, and psychographics of the respondents and the enumerators are significantly responsible for the improvement in literacy skills of the subtasks score of the treatment group, as well as the unimprovement values for the untreated groups as shown in the p-values for the treatment groups as shown in Tables 1–4. Issues like age, gender, accessibility to radio, living with parents, speaking of Hausa at home, family size, literacy levels of the parents/guardian, availability of books at home, and hunger were identified as indexes that help significantly shift the balance in favor of the treatment group.

Crucially, positive exposure is likely to significantly correlate positively with the various subtasks (Ho and Laflin 2009; Azevedo et al. 2020; Rasmitadila et al. 2020). For instance, a home with accessibility to radio where the parents have a positive attitude towards education and speak Hausa fluently is more likely to encourage their wards irrespective of gender and age to listen to the programs as broadcast on radio. This, though, does not portray a causal relationship as willingness to perform the subtasks or predisposition of the parents or other sociodemographic variables is not likely to result in improved performance. Hence, the need for further assessment and analysis in line with the ACADA model, so that appropriate actions can be taken.

In line with Azevedo et al. (2020) and Karam (2020), there are still some underlying variables that this study has not extensively explored. Additionally, the deployment of technology itself is likely to trigger other effects, as noted by (Ferguson et al. 2020). We agree with other approaches that suggest deploying other pedagogical strategies rather than the standard classroom activities (Johnson and Aragon 2003; Harris et al. 2010; Kristanto 2016) would trigger improvement in learning. Our study established that deploying local languages, in this case, Hausa and Fulfulde, is also important in promoting improved performance for various subtasks, as demonstrated by the marked differences in reading skills in Table 1 and Fig. 2. Other similar radio educational programs relish positive correlations in specific areas such as the improvement in learning geography (Caldis 2018), learning in a crisis environment (Driscoll 2010; Carlson and JBS International 2013; Education Development Center 2019), and even in normalcy (Patrick 2018, Obukoadata et al. 2020). However, all of these previous works were not age-biased as compared to this study which was specific to a class with an age range.

The results, also, show significantly better learning outcomes in the treatment groups encased within the sociodemographic profile of the respondents. This is noted in the attributes of the subtasks of letter and syllable identification, non-word reading, familiar word reading, and reading comprehensions which recorded higher mean p-values (see Tables 2 and 4) in comparison to the other groups and even higher than the initial SENSE baseline reading proficiency and performance assessment (AUN 2020).

Conclusion

Conclusively, we state that the mid-term SENSE-TRI evaluation had yielded positive results, as all five major subtasks of letter identification, syllable identification, non-word reading, familiar word reading, and reading comprehension recorded statistically significant higher scores for the treatment groups for all genders as compared to the comparison groups. As argued earlier, it is too preemptive to conclude that the SENSE-TRI program was exclusively responsible for the differences and that the program had peaked in terms of impact. But it would suffice to state that the core objectives of the SENSE-TRI are being met. First, the program provided a form of safe education opportunities with minimal resources for the subjects despite the ravages of COVID-19 and insurgency. Education was not only safe but also impactful as the data suggests better learning outcomes in the treatment groups. The delivery of conflict-sensitive instructions devoid of gender and religious biases was also a strong part of this program. In all, we could assert that the deployment of technology has helped deepen literacy skills among the participants. Nonetheless, such deepening does not at any level amplify the place of technology above physical contacts, but that blended and online learning are generally more effective than face-to-face instruction (Topping et al. 2022).

There is a need to interrogate the roles of the demographic differences between the treatment and comparison groups as indicated in their ages, parental attachment, and attitude toward learning at home, with or without food. We cannot also rule out the influences from peers during the lockdown, and the apathy towards the menace of the pandemic, which somewhat could either positively or negatively impact the program further. The expectations are that with the reopening of schools, and with the results so far achieved, the impact would be monumental, as well as help improve the basic literacy skills of letter identification, syllable identification, non-word reading, familiar word reading, and reading comprehension. The study revealed that all demographic characteristics showed statistically significant differences between the groups, indicating the potential for targeted interventions in specific areas. The program’s impact was enhanced significantly by the literacy abilities of parents who provided essential home support to the participants, underscoring the importance of parental involvement. The results also highlighted the crucial role of the ‘parent literacy’ variable in the program’s effectiveness. The study found that factors such as living with parents, family size, studying without food, and availability of books at home played a significant role in the treatment subjects’ response to the SENSE-TRI program. To maximize the program’s positive impact, it is recommended to encourage and further explore these factors.

These characteristics are very instructive in maintaining the level of impact of the SENSE-TRI program going forward. Strengthening the program implies that these other variables had to be strengthened as well. It also suggests that other areas where strong emphasis had been laid in the past should be downscaled for these new parameters. For instance, emphasis in the past had been placed on family sizes, food, and gender. More consideration should be given to inputs from parents and guardians, and subtle encouragement to make them cultivate the habit of reading, among others. When the children live with their parents, there is the tendency for the parents, who have interest in reading to also encourage them to read as well. In all, we contend from the results, that learners who participate in the SENSE-TRI program have better outcomes on the basic literacy skills of letter identification, syllable identification, non-word reading, familiar word reading, and reading comprehensions, and also in emergency situations such as COVID-19 pandemic lockdown and insurgency. Alternative ways of learning are inevitable but deploying basic technologies can enhance development along the ACADA model, for which we can integrate the TRI to form the ACADA-TRI model of learning during unusual periods. Overall, the SENSE-TRI shows a significant difference in performance between the treatment group and the comparison group.

The main weakness of this study is the lack of a baseline assessment of all learners, which would allow for comparing of learning gains between groups instead of just the literacy levels at a given point in time. A further weakness is the recourse to just a set of pupils, that is the grade 3 pupils, and the focus on just basic literacy skills. A more comprehensive study would be encouraged to look at the primary or basic educational structure in Nigeria as well as do a cross-sectional national survey so as to compare with more national results as published. Nonetheless, the study provides an expressive analysis of the application of the TRI model in situations such as insurgencies, counter-insurgencies, displacements, and even in normal situations that are hampered by poverty, hunger, environmental constraints, and other socioeconomic and political limitations. Therefore, suggestion for further studies would be to conduct a cross-national longitudinal survey, evaluating several other variables across several learning classes and ages, and also the workability of the SENSE-TRI model in the absence of the pandemic and insurgencies.

But we note also with deep concerns, that despite the improvement in the learning abilities of the children who participated in the SENSE-TRI program, consistent inaccessibility to physical classrooms in a fully digitalized era is most likely to further aggravate gender-based digital divide in developing countries, especially those in remote locations. Maity et al. (2022) had earlier noted these same concerns which we think needed to be addressed, as that could become another global issue. Another sordid implication for this work in deploying tech-based learning is the “sudden radical remodeling to the virtual mode of education” (Maity et al. 2022) which has the propensity to usher unprepared basic-level students into tech-based-culture-enabled-society unannounced. Whether this tectonic shift is good or bad should be the interest of further studies.

Specifically, the manuscript presents an affirmative evaluation of the SENSE-TRI program and offers statistically significant results, indicating its success in improving literacy skills among learners in the treatment group. The acknowledgment of confounding variables such as demographics and pandemic-related influences is effective. References to relevant studies provide supportive evidence for this conclusion, strengthening the argument further for the need to embrace more digital approaches to learning in the future. And, finally, the study, adds to the literature by exploring the impact of the SENSE-TRI program on learners’ literacy skills within a unique context, namely remote education during pandemic conditions in specific regions of Nigeria. This contributes to the broader academic discourse by providing insights from an underexplored perspective.