A study examined a reading recovery program in Great Bend, Kansas. Subjects were 70 first-grade students divided into a treatment group and a control group. In the treatment group, 19 were Caucasian, 6 were Hispanic, and 6 were Vietnamese or Afro-American. In the control group, 31 were Caucasian and 8 were Hispanic. Independent variables were: participation status, gender, ethnicity, and number of sessions. Dependent variables were: writing vocabulary and dictation. Four composite null hypotheses were tested at the .05 level of significance, employing single-factor analysis of covariance. A total of 8 comparisons were made employing a single-factor analysis of covariance. Of the 8 comparisons made, 1 was statistically significant at the .05 level. The statistically significant comparison was for the independent variable participation status and the dependent variable writing vocabulary. Results indicated that the mathematical assumption of homogeneity of regression was met for six of eight comparisons. The assumption was not met for the comparisons gender and writing vocabulary, and the ethnicity and writing vocabulary. Findings suggest that: (1) subjects who participated in the reading recovery program had a statistically higher mean writing vocabulary score than those who did not participate; (2) subjects in both groups had equal achievement in dictation; (3) males and females who participated did equally well; (4) subjects with different ethnicities who participated did equally well; and (5) subjects who participate in reading recovery for varying numbers of sessions do equally well. (Contains 4 tables of data and 82 references; 8 appendixes contain sample letters, lesson plans, and teacher documentation; training overview; documentation; and summary sheets.) (Author/CR)