CALIFORNIA STATE UNIVERSITY, BAKERSFIELD SCHOOL OF BUSINESS AND PUBLIC ADMINISTRATION Department of Public Policy and Administration PPA 415 – Research Methods in Public Administration Exercise 6 - Answers Question 1 2 3 4 5 Total Analysis 4 4 12 16 4 40 Interpretation 6 6 18 24 6 60 1. Do problem 9.4 in Healey (p. 222). Do calculations by hand (10 points – 4 analysis, 6 interpretation). 9.4 CJ The city of Shinbone has implemented two separate crime prevention programs. One involves a neighborhood watch program with citizens actively involved in crime prevention. The second involves officers patrolling the neighborhoods on food rather than in patrol cars. In terms of the percentage reduction in crimes reported to the police over a one-year period, were the programs successful? The results are for random samples of 18 neighborhoods drawn from the entire city. Step 1: Making the Assumptions and Meeting Test Requirements Model: Independent random samples Interval-ratio measurement of the dependent variable. Normally distributed populations Equal population variances Step 2: Stating the Null Hypothesis H0: 1 = μ2 = μ3 H1: At least one mean is significantly different Step 3: Selecting the Sampling Distribution and Establishing the Critical Region. Sampling distribution = F distribution = 0.05 Degrees of freedom (within) = N – k = 18 – 3 = 15 Degrees of freedom (between) = k – 1 = 3 – 1 = 2 F(critical) = 3.68 Step 4: Computing the Test Statistic Table 1. Analysis of Variance of Crime Prevention Strategy Outcomes Anova: Single Factor SUMMARY Groups Neighborhood Watch Foot Patrol No Program Count Sum 6 6 6 ANOVA Source of Variation Between Groups Within Groups SS 10208.44444 15973.33333 Total 26181.77778 Average Variance 80 13.33333333 986.6666667 -186 -31 800 144 24 1408 df MS F P-value F crit 2 5104.222222 4.793196995 0.024573848 3.682320344 15 1064.888889 17 Step 5: Making a Decision and Interpreting the Results of the Test F(obtained) of 4.79 is higher than F(critical) of 3.68. We can reject the null hypothesis that the three approaches to crime reduction have the same effect on crime. The data suggest that foot patrol has the strongest impact on reducing crime, whereas current practices have actually coincided with an increase in crime. 2. Do problem 9.7 in Healey (p. 223). Do problem in SPSS (10 points – 4 analysis, 6 interpretation). 9.7 GER Do older citizens lose interest in politics and current affairs? A brief quiz on recent headline stories was administered to random samples of respondents from each of four different age groups. Is there a significant difference? The data below represent numbers of correct responses. Table 2. The Influence of Age on Interest in Politics High Young Middle Retired School Adult Aged (15-18) (21-30) (30-55) (65+) 0 0 2 5 1 0 3 6 1 2 3 6 2 2 4 6 2 4 4 7 2 4 5 7 3 4 6 8 5 6 7 10 5 7 7 10 7 7 8 10 7 7 8 10 9 10 10 10 Step 1: Making the Assumptions and Meeting Test Requirements Model: Independent random samples Interval-ratio measurement of the dependent variable. Normally distributed populations Equal population variances Step 2: Stating the Null Hypothesis H0: 1 = μ2 = μ3 = μ4 H1: At least one mean is significantly different Step 3: Selecting the Sampling Distribution and Establishing the Critical Region. Sampling distribution = F distribution = 0.05 Degrees of freedom (within) = N – k = 48 – 4 = 44 Degrees of freedom (between) = k – 1 = 4 – 1 = 3 F(critical) = 2.816465827. Step 4: Computing the Test Statistic Table 3. Descriptive Statistics on Age and Political Interest Descriptives Correct res ponses on headline quiz N High School (15-18) Young Adult (21-30) Middle Aged (30-55) Retired (65+) Total 12 12 12 12 48 Mean 3.67 4.42 5.58 7.92 5.40 Std. Deviation 2.871 3.088 2.466 1.975 3.023 Std. Error .829 .892 .712 .570 .436 95% Confidence Interval for Mean Lower Bound Upper Bound 1.84 5.49 2.45 6.38 4.02 7.15 6.66 9.17 4.52 6.27 Minimum 0 0 2 5 0 Table 4. Analysis of Variance of Age and Political Interest ANOVA Correct res ponses on headline quiz Between Groups Within Groups Total Sum of Squares 124.063 305.417 429.479 df 3 44 47 Mean Square 41.354 6.941 F 5.958 Sig. .002 Maximum 9 10 10 10 10 Mean Number of Correct responses on headline quiz 10 9 8 7 6 5 4 3 2 1 0 High School (15-18) Young Adult (21-30) Middle Aged (30-55) Retired (65+) Age level Figure 1. Mean Political Interest by Age Step 5: Making a Decision and Interpreting the Results of the Test F(obtained) of 5.958 exceeds the F(critical) of 2.82. We can reject the null hypothesis that age has no influence on political interest. The researcher anticipated that political interest would decline with age; however, political interest appears to increase with age, although it is difficult to determine from a cross-sectional survey whether this is an aging or a generational effect. 3. Path-Goal Theory attempts to link leadership style to the characteristics of the subordinate and the type of work being conducted. Generally, the theory suggests the following patterns (30 points – 12 points analysis, 18 points interpretation). Table 5. Path-Goal Theory: How It Works Using the Path-Goal Leadership data set, determine whether the four leadership styles (directive, supportive, participative, participative, and achievement oriented) vary significantly by the type of organization (public, private, nonprofit) that the members of the class worked for. HINT: Be sure to use the variables with the full scores and not the ranked variables). Step 1: Making the Assumptions and Meeting Test Requirements Model: Independent random samples Interval-ratio measurement of the dependent variable. Normally distributed populations Equal population variances Step 2: Stating the Null Hypothesis H0: 1 = μ2 = μ3 H1: At least one mean is significantly different Step 3: Selecting the Sampling Distribution and Establishing the Critical Region. Sampling distribution = F distribution = 0.05 Degrees of freedom (within) = N – k = 39 – 3 = 36 Degrees of freedom (between) = k – 1 = 3 – 1 = 2 F(critical) = 3.259446306 Step 4: Computing the Test Statistic Table 6. Four Leadership Styles by Type of Organization Descriptives N Directive Style Supportive Style Participative Style Achievement-Oriented Style Public Private Nonprofit Total Public Private Nonprofit Total Public Private Nonprofit Total Public Private Nonprofit Total 17 17 5 39 17 17 5 39 17 17 5 39 17 17 5 Mean 28.65 28.94 29.00 28.82 28.18 29.94 32.00 29.44 25.35 24.59 25.80 25.08 27.12 27.18 26.60 Std. Deviation 4.256 4.841 4.359 4.418 3.167 3.152 1.581 3.218 2.957 3.203 2.775 3.003 4.923 4.447 2.510 Std. Error 1.032 1.174 1.949 .707 .768 .764 .707 .515 .717 .777 1.241 .481 1.194 1.079 1.122 39 27.08 4.385 .702 95% Confidence Interval for Mean Lower Bound Upper Bound 26.46 30.84 26.45 31.43 23.59 34.41 27.39 30.25 26.55 29.80 28.32 31.56 30.04 33.96 28.39 30.48 23.83 26.87 22.94 26.23 22.35 29.25 24.10 26.05 24.59 29.65 24.89 29.46 23.48 29.72 25.66 Minimum 20 18 22 18 22 23 30 22 18 19 21 18 18 15 23 Maximum 35 35 33 35 34 35 34 35 29 31 28 31 35 33 29 15 35 28.50 Table 7. Analysis of Variance of Path-Goal Leadership Style by Type of Organization ANOVA Directive Style Supportive Style Participative Style Between Groups Within Groups Total Between Groups Within Groups Total Between Groups Within Groups Total Between Groups Within Groups Total 35 35 30 30 25 Mean of Supportive Style Mean of Directive Style Achievement-Oriented Style Sum of Squares .920 740.824 741.744 64.178 329.412 393.590 7.969 334.800 342.769 1.334 729.435 730.769 20 15 10 5 df 2 36 38 2 36 38 2 36 38 2 36 38 Mean Square .460 20.578 F .022 Sig. .978 32.089 9.150 3.507 .041 3.985 9.300 .428 .655 .667 20.262 .033 .968 25 20 15 10 5 0 0 Public Private Nonprofit Type of Organization Figure 2. Directive Style by Type of Organization Public Private Nonprofit Type of Organization Figure 3. Supportive Style of Organization 35 30 30 Mean of Achievement-Oriented Style Mean of Participative Style 35 25 20 15 10 5 25 20 15 10 5 0 0 Public Private Nonprofit Type of Organization Figure 4. Participative Style by Type of Organization Public Private Nonprofit Type of Organization Figure 5. Achievement-Oriented Style by Type of Organization Step 5: Making a Decision and Interpreting the Results of the Test For three of the analyses (Directive, Participative, and Achievement-Oriented Styles), Fobtained is less than F-critical (.022, .438, .033 versus 3.26). We cannot reject the null hypothesis that type of organization has no impact on across these three leadership styles. For the supportive style, F-obtained of 3.507 exceeds the F-critical of 3.26. We can reject the null hypothesis that organization has no impact on supportive leadership. In fact, students in the leadership class who served in nonprofit organizations had higher average supportive leadership scores than private sector students. Public sector students had the lowest averages, about 12 percent less than nonprofit students. 4. The second round of the CSUB Policy Delphi asked faculty, administrators, students, staff, and community members to rank ten learning outcomes (critical speaking, critical reading, ethical framework, working independently, critical writing, technology applications to problem solving, application of discipline to realworld, critical thinking, diversity and cultural understanding, basic understanding of a discipline [Speaking to Undertanding]) from first to tenth most important. The most important was given a score of ten and the least important a score of one. Does the mean ranking of each learning outcome vary significantly by the respondent’s relationship to the university? Use the .10 level of significance (alpha). HINT: Use the recoded relationship variable (30 points – 12 points analysis, 18 interpretation). Step 1: Making the Assumptions and Meeting Test Requirements Model: Independent random samples Interval-ratio measurement of the dependent variable. Normally distributed populations Equal population variances Step 2: Stating the Null Hypothesis H0: 1 = μ2 = μ3 H1: At least one mean is significantly different Step 3: Selecting the Sampling Distribution and Establishing the Critical Region. Sampling distribution = F distribution = 0.10 Dfw (critical speaking) = N – k = 237 – 3 = 234 Dfw (critical reading) = N – k = 238 – 3 = 235 Dfw (ethical framework) = N – k = 233 – 3 = 230 Dfw (working independently) = N – k = 232 – 3 = 229 Dfw (critical writing) = N – k = 235 – 3 = 232 Dfw (technology applications to problem solving) = N – k = 237 – 3 = 234 Dfw (application of discipline to real world) = N – k = 237 – 3 = 234 Dfw (critical thinking) = N – k = 241 – 3 = 238 Dfw (diversity and cultural understanding) = N – k = 237 – 3 = 234 Dfw (basic understanding of a discipline) = N – k = 240 – 3 = 237 Dfb = k – 1 = 3 – 1 = 2 F-critical (critical speaking) = 2.325 F-critical (critical reading) = 2.325 F-critical (ethical framework) = 2.326 F-critical (working independently) = 2.326 F-critical (critical writing) = 2.326 F-critical (technology applications to problem solving) = 2.325 F-critical (application of discipline to real world) = 2.325 F-critical (critical thinking) = 2.325 F-critical (diversity and cultural understanding) = 2.325 F-critical (basic understanding of a discipline) = 2.325 Step 4: Computing the Test Statistic Table 8. Student Learning Outcome Ranking by Role at University Descriptives N Critical s peaking. Critical reading. Ethical framework. Work Independently Critical writing. Technology applications to problem s olving. Application of dis cipline to real-world. Critical thinking. Divers ity and cultural understanding. Bas ic unders tanding of a discipline. Faculty, staff, or adminis trator Student Alumni or community member Total Faculty, staff, or adminis trator Student Alumni or community member Total Faculty, staff, or adminis trator Student Alumni or community member Total Faculty, staff, or adminis trator Student Alumni or community member Total Faculty, staff, or adminis trator Student Alumni or community member Total Faculty, staff, or adminis trator Student Alumni or community member Total Faculty, staff, or adminis trator Student Alumni or community member Total Faculty, staff, or adminis trator Student Alumni or community member Total Faculty, staff, or adminis trator Student Alumni or community member Total Faculty, staff, or adminis trator Student Alumni or community member Total 58 118 61 237 58 119 61 238 58 114 61 233 57 115 60 232 57 119 59 235 55 121 61 237 58 118 61 237 58 120 63 241 58 119 60 237 58 120 62 240 Mean 4.88 5.42 5.95 5.42 6.81 6.18 6.57 6.43 6.38 5.32 5.89 5.73 4.42 5.01 4.27 4.67 6.25 6.18 6.17 6.20 3.36 3.64 3.97 3.66 5.43 6.52 6.41 6.22 8.76 8.23 8.22 8.36 3.71 3.97 3.00 3.66 5.00 5.15 4.98 5.07 Std. Deviation 2.541 2.684 2.533 2.628 2.131 2.466 2.239 2.337 2.621 2.652 2.409 2.610 2.291 2.799 2.524 2.623 2.132 2.361 2.245 2.269 2.189 2.221 2.429 2.269 2.555 2.819 2.710 2.755 1.848 2.028 2.413 2.099 2.656 2.911 2.636 2.799 3.340 3.020 3.257 3.149 Std. Error .334 .247 .324 .171 .280 .226 .287 .151 .344 .248 .308 .171 .303 .261 .326 .172 .282 .216 .292 .148 .295 .202 .311 .147 .336 .259 .347 .179 .243 .185 .304 .135 .349 .267 .340 .182 .439 .276 .414 .203 Table 9. Analysis of Variance of Student Learning Outcome Rankings by Role at University ANOVA Critical s peaking. Critical reading. Ethical framework. Work Independently Critical writing. Technology applications to problem s olving. Application of dis cipline to real-world. Critical thinking. Divers ity and cultural unders tanding. Bas ic unders tanding of a dis cipline. Between Groups Within Groups Total Between Groups Within Groups Total Between Groups Within Groups Total Between Groups Within Groups Total Between Groups Within Groups Total Between Groups Within Groups Total Between Groups Within Groups Total Between Groups Within Groups Total Between Groups Within Groups Total Between Groups Within Groups Total Sum of Squares 34.146 1595.660 1629.806 17.298 1277.126 1294.424 45.482 1534.483 1579.966 26.484 1562.619 1589.103 .196 1204.799 1204.996 10.655 1204.662 1215.316 48.703 1742.444 1791.148 12.335 1044.976 1057.311 37.434 1811.883 1849.316 1.512 2368.284 2369.796 df 2 234 236 2 235 237 2 230 232 2 229 231 2 232 234 2 234 236 2 234 236 2 238 240 2 234 236 2 237 239 Mean Square 17.073 6.819 F 2.504 Sig. .084 8.649 5.435 1.592 .206 22.741 6.672 3.409 .035 13.242 6.824 1.941 .146 .098 5.193 .019 .981 5.327 5.148 1.035 .357 24.352 7.446 3.270 .040 6.167 4.391 1.405 .247 18.717 7.743 2.417 .091 .756 9.993 .076 .927 Step 5: Making a Decision and Interpreting the Results of the Test For six of the analyses (critical reading, working independently, critical writing, technology applications, critical thinking, diversity, and disciplinary understanding), Fobtained is less than F-critical (.019 to 1.942 versus 2.33). We cannot reject the null hypothesis that role at the university has no impact on these six learning outcomes. For four of the learning outcomes (critical speaking, ethical framework, application of discipline to the real world, and diversity), F-obtained of 2.417 to 3.409 exceeds the Fcritical of 2.33. We can reject the null hypothesis that university role has no impact on these four learning outcomes. In general, alumni and community members rate critical speaking highest, whereas faculty members rate it lowest. Faculty members rate ethics highest, and students rate it lowest. Students rate application of disciplines to the real world highest and faculty rate it lowest. Students also rate diversity and cultural understanding highest, while community members rate it lowest. 5. Several authors have suggested that some regions of the country are more likely to receive disaster declarations than other regions of the country. Does the probability of a major disaster declaration (if SBA declarations are treated as turndowns [ActionType2]) vary significantly by FEMA region (10 points – 4 points analysis, 6 points interpretation)? Step 1: Making the Assumptions and Meeting Test Requirements Model: Independent random samples Interval-ratio measurement of the dependent variable. Normally distributed populations Equal population variances Step 2: Stating the Null Hypothesis H0: 1 = μ2 = μ3 = μ4 = 5 = μ6 = μ7 = μ8 = μ9 = μ10 H1: At least one mean is significantly different Step 3: Selecting the Sampling Distribution and Establishing the Critical Region. Sampling distribution = F distribution = 0.05 Degrees of freedom (within) = N – k = 539 – 10 = 529 Degrees of freedom (between) = k – 1 = 10 – 1 = 9 F(critical) = 1.897570696. Step 4: Computing the Test Statistic Table 10. Disaster Declaration Rates by FEMA Region, 1953 - 1973 Descriptives Presidential Disaster Decision (SBA as Turndowns) N 1 Connecticut, Maine, Mas s achusetts , New Hampshire, Rhode Is land, Vermont 2 New Jersey, New York, Puerto Rico, and the Virgin Is lands 3 Delaware, Dis trict of Columbia, Maryland, Penns ylvania, Virginia and W. Virginia 4 Alabama, Florida, Georgia, Kentucky, Miss is sippi, N. Carolina, S. Carolina and Tennes see 5 Illinois , Indiana, Michigan, Minnesota, Ohio and Wisconsin 6 Arkansas, Louisiana, New Mexico, Oklahoma and Texas 7 Iowa, Kansas, Miss ouri and Nebraska 8 Colorado, Montana, N. Dakota, S. Dakota, Utah and Wyoming 9 Arizona, California, Hawaii, Nevada, American Samoa, Guam, Northern Mariana Is lands, Marshall Is lands, Micrones ia 10 Alas ka, Idaho, Oregon and Was hington Total 23 20 43 104 72 87 52 36 66 36 539 Mean .826 .850 .791 .548 .639 .667 .692 .639 .606 .833 .668 Std. Deviation .3876 .3663 .4116 .5001 .4837 .4741 .4660 .4871 .4924 .3780 .4714 Table 11. ANOVA of Disaster Declaration Rates by FEMA Region, 1953 - 1973 ANOVA Presidential Disaster Decision (SBA as Turndowns ) Between Groups Within Groups Total Sum of Squares 4.740 114.815 119.555 df 9 529 538 Mean Square .527 .217 F 2.427 Sig. .010 Std. Error .0808 .0819 .0628 .0490 .0570 .0508 .0646 .0812 .0606 .0630 .0203 Mean of Presidential Disaster Decision (SBA as Turndowns) 1.0 0.9 0.8 0.7 0.6 0.5 0.4 0.3 0.2 0.1 0.0 Connecticut, New Delaware, Alabama, Illinois, Maine, Jersey, District of Florida, Indiana, Massachus New York, Columbia, Georgia, Michigan, etts, New Puerto Rico, Maryland, Kentucky, Minnesota, Hampshire, and the Pennsylvani Mississippi, Ohio and Rhode Virgin a, Virginia N. Carolina, Wisconsin Island, Islands and W. S. Carolina Vermont Virginia and Tennessee Arkansas, Louisiana, New Mexico, Oklahoma and Texas Iowa, Kansas, Missouri and Nebraska Colorado, Arizona, Montana, N. California, Dakota, S. Hawaii, Dakota, Nevada, Utah and American Wyoming Samoa, Guam, Northern Mariana Islands, Marshall Islands, Micronesia Alaska, Idaho, Oregon and Washington FEMA Region Figure 6. Disaster Declaration Rates by FEMA Region, 1953 – 1973 Step 5: Making a Decision and Interpreting the Results of the Test F(obtained) of 2.427 is greater than F(critical) of 1.898. We can reject the null hypothesis that region does not influence presidential disaster decisions. Generally, during the period 1953 to 1973, FEMA regions 1, 2, 3, and 10 had higher approval rates (79% to 85%) and FEMA region 4 had the lowest (55%). The remaining five regions varied from 61 to 69 percent approval, very close to the overall average of 67 percent.