For whom does the Good Behavior Game work?

For whom does the Good Behavior Game work?

By Claire Chuter, Johns Hopkins University

There is ample evidence that social-emotional learning programs support behavioral and academic outcomes in students. However, few studies have looked at the “who” and “why” that make these programs work. In this study, implementation variability and participant risk status were examined as predictors of disruptive behavior outcomes. In this large randomized experiment, seventy-seven English primary schools (N = 3,084 children, aged 6–7) were assigned to either receive the Good Behavior Game or to continue with business-as-usual. The Good Behavior Game is a universal behavior management intervention that encourages students to monitor their behavior in return for tangible rewards. 

Due to the clustered nature of the data, hierarchical linear models were fitted to the data. This study used intent-to-treat as well as complier average causal effects samples in order to compare findings between the two. Interestingly, intent-to-treat analysis found no discernible impact of the intervention on children’s disruptive behavior. However, complier average causal effect estimation (using dosage as a compliance marker) found a large, statistically significant intervention effect (d = -1.35) among compliers who spent at least 1,030 minutes using the Good Behavior Game over two years.

Moreover, the effect among compliers varied by students’ risk factors. Students with a high number of behavioral risk factors improved more with increased exposure (β = .41, p < .001) and less with decreased exposure than students not at risk (β = .22, p < .001). This study highlights the value in the whole class playing games such as the Good Behavior Game, even solely for the benefit of a few students who are most at risk.

Understanding the dynamics of dosage response: A nonlinear meta-analysis of recent reading interventions

Understanding the dynamics of dosage response: A nonlinear meta-analysis of recent reading interventions

By José L. Arco-Tirado, Faculty of Education, University of Granada (Spain)

A recent meta-analysis published in the Review of Educational Research intends to identify and understand the intervention characteristics associated with the largest reading effect sizes.

To support students’ reading outcomes, current models of intervention delivery have utilized multi-tiered systems of support (MTSS), also referred to as Response to Intervention (RtI). Within current MTSS and RtI frameworks, intervention levels are organized around three tiers. Tier 1 consists of delivering general education classroom instruction. Tier 2 consists of small group or 1:1 tutoring. Tier 3 consists of 1:1 instruction, with higher dosage and personalization.

However, despite the promise of early reading interventions, about 18% to 55% of K–3 students with reading disabilities (SWRD) under Tier 2 intervention have continued to struggle in reading. Therefore, it remains critical to better understand how and when to intensify reading interventions

Since linear models in intervention research and meta-analyses have been unable to substantiate the claim that a larger dosage (i.e., more hours of intervention) produces significantly larger effect sizes than interventions with less dosage (i.e., fewer hours of intervention) and based on the observed nonlinear association between reading instruction (i.e., hours of instruction) and reading outcomes for Grade K-3 students with reading difficulties (K-3 SWRD), the authors conducted a nonlinear meta-analysis to investigate the nonlinear reported effect sizes through modelling the maximum effect and optimal dosage to respond to the following research questions: (a) what is the optimal dosage and maximum predicted effect size from reading interventions for K–3 SWRD? and (b) when investigating outcome type, intervention components, group size, or norm-referenced moderators, to what extent did the optimal dosage and maximum predicted effect size vary from the overall optimal dosage and maximum predicted effect size presented in the first research question? Results suggest that reading intervention effect sizes, relative to a comparison condition, increase until approximately 40 hours of small group K–3 SWRD reading instruction. After this point, effect sizes tend to decline. For students who have inadequately responded to small group reading instruction, we also identified 1:1 groupings as a possible method to increase student outcomes after the 40-hour time point is reached.

Effectiveness of volunteer tutoring

Effectiveness of volunteer tutoring

By Justin Hill, Johns Hopkins University

Carrie E. Markovitz and colleagues recently reported on a replication and expansion of a previous randomized controlled trial  focused on volunteer tutoring in reading for at-risk early elementary school students. The current study focuses on the effectiveness of the Minnesota Reading Corps and the Wisconsin Reading Corps, which are both programs within AmeriCorps. The initial 2014 study focused solely on Minnesota and was limited in its ability to assess impacts for second and third grade students. The authors suggest the current study is useful because aspects of the tutoring programs have changed, they are now evaluating the effects of tutoring in two separate programs, and they are now able to have a longer evaluation of the effects on second and third grade students. The current study used a matched-pairs design in which students were matched based upon their baseline fall test scores, and then one student was assigned to the control group while the other was assigned to the tutoring program. The Minnesota portion of the study utilized 60 kindergarten students, 160 first-grade students, 190 second-grade students, and 212 third-grade students while the Wisconsin portion enrolled 64 kindergarten students and 112 first-grade students.

In Minnesota, kindergarten students in the tutoring program for one semester identified 10.9 more letter sounds within one minute than students in the control group (ES = + 0.85, p = .01). First grade students in the tutoring program for one semester identified 16.3 more letter sounds within one minute than students in the control group (ES = + 0.81, p < .001) and read 13.3 more words aloud than students in the control group (ES = + 0.61, p = .02). Finally, second and third grade students in the tutoring program for one year read 6.4 more words aloud in one minute than students in the control group (ES = + 0.28, p < .01).

In Wisconsin, kindergarten students in the tutoring program for one semester identified 6.5 more letter sounds within one minute than students in the control group (ES = + 0.55, p = .04). First grade students in the tutoring program for one semester identified 8.7 more letter sounds within one minute than students in the control group (ES = + 0.46, p < .01). In both states, the strongest effects were noted for younger children. Kindergarten students in the tutoring program in both states surpassed the benchmark achievement level while students in the control group remained behind grade level. Despite also making progress, older students in the tutoring program were not able to achieve their grade-level benchmark scores. This study demonstrates the effectiveness of structured volunteer tutoring in reading, especially for younger students.

Relations among phonological processing skills and mathematics

Relations among phonological processing skills and mathematics

By Winnie Tam, Centre for University and School Partnership, The Chinese University of Hong Kong

A recent meta-analysis was published that investigated the association between phonological processing skills and mathematics in children. Phonological processing refers to the use of the sound structure of language to manage written and oral information and consists of three components, namely, phonological awareness (PA), rapid automatized naming (RAN), and phonological memory (PM). Phonological awareness refers to awareness of the sound structure of language. RAN signifies the rate of access to phonological information in long-term memory, and it is usually measured by how fast an individual can name symbol stimuli (e.g., colors, letters, digits). In contrast, PM comprises short-term phonological storage and a rehearsal process that maintains decaying information. It is usually accessed by the number of stimuli to be recalled within a string of words, letters, or digits.

A total of 94 studies (135 unique samples, 826 effect sizes) were examined. To be included for analysis, studies had to focus on participants in kindergarten or primary school. Effect sizes and Pearson’s correlation were collected between the phonological processing measures and the mathematics outcomes, as well as children’s characteristics (e.g., age, grade). Results of the meta-analysis are shown below.

  • In general, the relation between phonological processing and mathematics (r = +0.33) was significant.
  • For subgroups, significant relations were also found among kindergarteners (K1-K3, r =+0.36), junior primary school children (G1-G3, r =+0.32), and senior primary school students (G4-G6, r =+0.29).
  • PA (r = +0.38) and RAN (r = +0.35) showed stronger correlation with mathematics skills than PM (r =+0.28) did.
  • Mean age was a significant moderator: younger children demonstrated a stronger relation between phonological processing skills and mathematics.
  • Moreover, the association of RAN with mathematics was significantly stronger among kindergarteners (K1-K3) than senior primary school students (G4-G6).
  • PA and PM showed a stronger association with mathematics accuracy than with mathematics fluency, whereas RAN was more strongly related to mathematics fluency than to mathematics accuracy.

The findings suggest that early PA and RAN abilities may be a good indicator of mathematics performance. Early training in recognition and manipulation of speech sounds as well as phonological access to symbols may facilitate early mathematics learning.