どのSNSを使う人が満足してるか
2019年度調査実習分析例
石田淳
目的
エンタメ班のデータをもとに、どのようなSNSを使うが各種満足感が高いのかを分析する。
データ成型
library(tidyverse)## -- Attaching packages -------------------------------------- tidyverse 1.2.1 --## √ ggplot2 3.2.1 √ purrr 0.3.2
## √ tibble 2.1.3 √ dplyr 0.8.3
## √ tidyr 1.0.0 √ stringr 1.4.0
## √ readr 1.3.1 √ forcats 0.4.0## -- Conflicts ----------------------------------------- tidyverse_conflicts() --
## x dplyr::filter() masks stats::filter()
## x dplyr::lag() masks stats::lag()read_csv("9259000003_Rawdata_2.csv") %>% mutate(
SEX2 = recode_factor(SEX,`1` = "男",
`2` = "女",
.default = NA_character_),
eduyear = recode(q3,
`1` = 9,
`2` = 12,
`3` = 14,
`4` = 12,
`5` = 14,
`6` = 16,
`7` = 18,
.default = NA_real_),
work_manzoku = na_if(q20s1, 6),
work_manzoku = 6 - work_manzoku,
income_manzoku = na_if(q20s2, 6),
income_manzoku = 6 - income_manzoku,
relation_manzoku = na_if(q20s3, 6),
relation_manzoku = 6 - relation_manzoku,
life_manzoku = na_if(q20s4, 6),
life_manzoku = 6 - life_manzoku,
SNS = recode_factor(q23,`1` = "Twitter",
`2` = "Istagram",
`3` = "Facebook",
.default = NA_character_)
) -> data## Parsed with column specification:
## cols(
## .default = col_double(),
## MID = col_character(),
## DAY = col_character(),
## q5t19 = col_character(),
## q5t19_coded = col_character(),
## q6t7 = col_character(),
## q6t7_coded = col_character(),
## q11_coded = col_character(),
## q11t11 = col_character(),
## q11t11_coded = col_character(),
## q12_coded = col_character(),
## q12t7 = col_character(),
## q12t7_coded = col_character(),
## q17t9 = col_character(),
## q18_coded = col_character(),
## q18t8 = col_character(),
## q18t8_coded = col_character(),
## q31t6 = col_character(),
## q32t1 = col_character()
## )## See spec(...) for full column specifications.基礎分析
どのSNSを一番よく使うか
barplot(table(data$SNS,useNA = "always"))
男女別の違い
prop.table(table(data$SEX2,data$SNS),1)##
## Twitter Istagram Facebook
## 男 0.60629921 0.27952756 0.11417323
## 女 0.43911439 0.46863469 0.09225092mosaicplot(~ SEX2 + SNS, data = data)
SNSによる各満足感の違い
boxplot(work_manzoku ~ SNS, data = data)
boxplot(income_manzoku ~ SNS, data = data)
boxplot(relation_manzoku ~ SNS, data = data)
boxplot(life_manzoku ~ SNS, data = data)
棒グラフで比較
data %>% group_by(SNS) %>%
summarise(work = mean(work_manzoku,na.rm = TRUE),
income = mean(income_manzoku,na.rm = TRUE),
relation = mean(relation_manzoku,na.rm = TRUE),
life = mean(life_manzoku,na.rm = TRUE)) %>%
filter(SNS != "NA") %>%
gather(key = satis, value = value, -SNS) %>%
ggplot(aes(x=satis,y=value,fill = SNS)) +
geom_bar(stat="identity",position = "dodge")## Warning: Factor `SNS` contains implicit NA, consider using
## `forcats::fct_explicit_na`
SNSの違いによる満足感の違いに有意な効果はあるか
統制変数として年齢・性別・教育年数をぶち込む。
summary(lm(work_manzoku ~ AGE + SEX2 + eduyear + SNS, data = data))##
## Call:
## lm(formula = work_manzoku ~ AGE + SEX2 + eduyear + SNS, data = data)
##
## Residuals:
## Min 1Q Median 3Q Max
## -2.49127 -0.87374 0.06846 0.77304 2.08457
##
## Coefficients:
## Estimate Std. Error t value Pr(>|t|)
## (Intercept) 2.117562 0.497625 4.255 2.52e-05 ***
## AGE -0.005372 0.009845 -0.546 0.58560
## SEX2女 -0.074969 0.111078 -0.675 0.50006
## eduyear 0.085270 0.026511 3.216 0.00139 **
## SNSIstagram 0.158870 0.118430 1.341 0.18041
## SNSFacebook 0.137625 0.185762 0.741 0.45914
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## Residual standard error: 1.182 on 472 degrees of freedom
## (186 observations deleted due to missingness)
## Multiple R-squared: 0.02645, Adjusted R-squared: 0.01614
## F-statistic: 2.565 on 5 and 472 DF, p-value: 0.02643summary(lm(income_manzoku ~ AGE + SEX2 + eduyear + SNS, data = data))##
## Call:
## lm(formula = income_manzoku ~ AGE + SEX2 + eduyear + SNS, data = data)
##
## Residuals:
## Min 1Q Median 3Q Max
## -2.02476 -0.96332 0.05163 1.03344 2.67882
##
## Coefficients:
## Estimate Std. Error t value Pr(>|t|)
## (Intercept) 1.215594 0.505399 2.405 0.016530 *
## AGE -0.003561 0.010104 -0.352 0.724662
## SEX2女 -0.052311 0.115322 -0.454 0.650312
## eduyear 0.101925 0.026964 3.780 0.000176 ***
## SNSIstagram 0.192782 0.122816 1.570 0.117129
## SNSFacebook 0.285196 0.192410 1.482 0.138916
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## Residual standard error: 1.252 on 494 degrees of freedom
## (164 observations deleted due to missingness)
## Multiple R-squared: 0.03565, Adjusted R-squared: 0.02589
## F-statistic: 3.653 on 5 and 494 DF, p-value: 0.002962summary(lm(relation_manzoku ~ AGE + SEX2 + eduyear + SNS, data = data))##
## Call:
## lm(formula = relation_manzoku ~ AGE + SEX2 + eduyear + SNS, data = data)
##
## Residuals:
## Min 1Q Median 3Q Max
## -2.65209 -0.71267 0.01017 0.78686 2.36154
##
## Coefficients:
## Estimate Std. Error t value Pr(>|t|)
## (Intercept) 1.884443 0.470305 4.007 7.08e-05 ***
## AGE -0.003998 0.009527 -0.420 0.674948
## SEX2女 0.114801 0.109254 1.051 0.293864
## eduyear 0.093995 0.025439 3.695 0.000244 ***
## SNSIstagram 0.010552 0.116494 0.091 0.927864
## SNSFacebook 0.292832 0.183845 1.593 0.111827
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## Residual standard error: 1.2 on 505 degrees of freedom
## (153 observations deleted due to missingness)
## Multiple R-squared: 0.03362, Adjusted R-squared: 0.02406
## F-statistic: 3.514 on 5 and 505 DF, p-value: 0.003926summary(lm(life_manzoku ~ AGE + SEX2 + eduyear + SNS, data = data))##
## Call:
## lm(formula = life_manzoku ~ AGE + SEX2 + eduyear + SNS, data = data)
##
## Residuals:
## Min 1Q Median 3Q Max
## -2.57717 -0.57563 0.04835 0.74857 2.04964
##
## Coefficients:
## Estimate Std. Error t value Pr(>|t|)
## (Intercept) 1.7980157 0.4427605 4.061 5.66e-05 ***
## AGE -0.0005142 0.0089562 -0.057 0.954
## SEX2女 0.1043824 0.1028951 1.014 0.311
## eduyear 0.0974430 0.0239350 4.071 5.42e-05 ***
## SNSIstagram 0.1285376 0.1095632 1.173 0.241
## SNSFacebook 0.1263852 0.1736365 0.728 0.467
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## Residual standard error: 1.135 on 510 degrees of freedom
## (148 observations deleted due to missingness)
## Multiple R-squared: 0.03459, Adjusted R-squared: 0.02513
## F-statistic: 3.655 on 5 and 510 DF, p-value: 0.002937結論から言えば有意差はない。