Add PHIA surveys

[athowes]

• Could add more surveys
• There are sexual behaviour (SB) questions in MICS 4 and 5 but not 2 and 3
• It seems like PHIA is less standardised than MICS (?) but I looked at ZIMPHIA 2015-2016 and that had the sexual behaviour module

[athowes]

• Add more explanation of the DHS variables to the manuscript
• PHIA tend to be in well surveyed countries
• Oli may have conducted some sex behaviour extraction of PHIA:
  • Step 1: Individual data into categories
  • Step 2: aggregated categories (same as any other survey)
• MICS don't have any cluster coordinates, so districts would have to be randomly imputed. Suggested not to include right now
• Create figure and table about about all surveys used: country, survey, sample size, extra details
• GATHER compliance
• calc_survey_indicators is the function used everywhere to calculate proportions by stratification

[athowes]

The relevant PHIA surveys are:

• LSO2016PHIA
• MWI2015PHIA
• NAM2017PHIA
• SWZ2016PHIA
• TZA2016PHIA
• UGA2016PHIA
• ZMB2016PHIA
• CMR2017PHIA

[athowes]

The steps needed I think are:

1. Access all the prior objects like survey_meta survey_regions
2. Transform them to have the same structure as DHS surveys (following this example)
3. See if they can be merged together with DHS versions (e.g. a bind_rows) to use one call of calc_survey_indicators

[athowes]

Want to replicate this section of create_sexbehav_dhs (including the Adam adjustments at the end)

[athowes]

To-do on this tomorrow:

• [x] Rewrite the previous commit into a function
• [x] Create survey_sexbehav object in lso_survey_phia, save as artefact
• [x] Import artefact into lso_survey_behav and attempt to run calc_survey_indicators on the PHIA data, then merge the results with that of running calc_survey_indicators for the DHS data (alternatively attempt to merge the PHIA data with the DHS data and run calc_survey_indicators once)
• [ ] Go back over DHS coding of the categories and see if improvements can be made, or more importantly perhaps make sure that the approach is standard across types of survey #44

[athowes]

Based on the below call, it looks like everyone who is in not exactly one category is in none of them. Hopeful that this can be fixed.

> survey_sexbehav %>%
+   mutate(
+     r_tot = nosex12m + sexcohab + sexnonreg + sexpaid12m
+   ) %>%
+   filter(
+     r_tot != 1
+   ) %>%
+   summary()
  survey_id         individual_id          sex12m     nosex12m    sexcohab   sexnonreg   sexpaid12m sexnonregplus     r_tot  
 Length:119         Length:119         Min.   :1   Min.   :0   Min.   :0   Min.   :0   Min.   :0    Min.   :0     Min.   :0  
 Class :character   Class :character   1st Qu.:1   1st Qu.:0   1st Qu.:0   1st Qu.:0   1st Qu.:0    1st Qu.:0     1st Qu.:0  
 Mode  :character   Mode  :character   Median :1   Median :0   Median :0   Median :0   Median :0    Median :0     Median :0  
                                       Mean   :1   Mean   :0   Mean   :0   Mean   :0   Mean   :0    Mean   :0     Mean   :0  
                                       3rd Qu.:1   3rd Qu.:0   3rd Qu.:0   3rd Qu.:0   3rd Qu.:0    3rd Qu.:0     3rd Qu.:0  
                                       Max.   :1   Max.   :0   Max.   :0   Max.   :0   Max.   :0    Max.   :0     Max.   :0

[athowes]

Let NIQ be "not in questionnaire". Here are the list of deviations away from LEPHIA, probably need to be dealt with a bit manually:

Survey	Variable	Issue	Fixed?
ZimPHIA	analsxever	sex is defined as either anal or vaginal	Y: removed check for anal sex
ZimPHIA	lifetimesex, lifetimesexdk	NIQ	Y: not using
ZimPHIA	part12modkr	called part12monumdk instead	Y
ZamPHIA	analsxever	called ansxfrstage instead	Y: removed check for anal sex
ZamPHIA	lifetimesex, lifetimesexdk	called partlifetm, partlifetmdk instead	Y: not using
ZamPHIA	part12monum	called part12mo instead	Y
UPHIA	analsxever	sex is defined as either anal or vaginal	Y: removed check for anal sex
UPHIA	sellsx12mo	NIQ	Y
UPHIA	buysx12mo	NIQ	Y
THIS	lifetimesex, lifetimesexdk	NIQ	Y: not using
THIS	partlastsxtimed1, partlastsxtimed2, partlastsxtimed3	NIQ	Y: not using
SHIMS2	sellsx12mo	in questionnaire but not in our ind extract (why?)	Y (email pending)
NAMPHIA	analsxever	has kindsex_a, kindsex_b, ... instead. Also skips section if no vaginal sex	Y: removed check for anal sex
NAMPHIA	partlastsxtimed1, partlastsxtimed2, partlastsxtimed3	NIQ	Y
MPHIA	analsxever	called ansxyn instead	Y: removed check for anal sex
MPHIA	lifetimesex, lifetimesexdk	NIQ	Y: not using
MPHIA	part12modkr	called part12monumdk instead	Y
CAMPHIA	analsxever	ever had sex defined via evrhdsx "Sex can be when a penis enters a vagina, when a penis enters an anus, or when a person puts his/her mouth on the penis or vagina of another person."	Y: removed check for anal sex
CAMPHIA	partlastsxtimed1, partlastsxtimed2, partlastsxtimed3	NIQ	Y: not using
CAMPHIA	sellsx12mo	NIQ	Y

[athowes]

Running through the combination pipeline for ZWE as example:

• [x] zwe_survey_phia run and commit
• [ ] zwe_survey_sexbehav run and commit

[athowes]

• [x] Fix remaining issue with this table add payments for sex loop in last three partners to sexpaid12m.
• [x] Move those functions out to naomi.utils
• [ ] Rerun through all of the surveys to see if there are any remaining issues

[athowes]

aaa_survey_behav failed on the following. Find way to fix:

• [x] CMR, see commit below.
• [ ] TZA:

  Error in assign_dhs_cluster_areas(survey_clusters, survey_region_areas) : 
  Survey clusters have survey_region_id not contained in survey_regions:
  survey_id survey_region_id cluster_id
  1  TZA2007AIS               54          4
  2  TZA2007AIS               53         11
  3  TZA2007AIS               51         13
  4  TZA2007AIS               51         16
  5  TZA2007AIS               52         18
  6  TZA2007AIS               53         26
  7  TZA2007AIS               54         27
  8  TZA2007AIS               54         34
  9  TZA2007AIS               53         41
  10 TZA2007AIS               55         42
  11 TZA2007AIS               53         46
  12 TZA2007AIS               55         47
  13 TZA2007AIS               51         49
  14 TZA2007AIS               55         51
  15 TZA2007AIS               55         54
  16 TZA2007AIS               53         61
  17 TZA2007AIS               52         68
  18 TZA2007AIS               52         78
  19 TZA2007AIS               51         88
  20 TZA2007AIS               52         91
  21 TZA2007AIS               52    

  Kinh says that he has seen this error before and it was fixed by switching to a development branch of naomi.utils.

[athowes]

The PHIA part of this is done, so closing and adding new issue relating to adding new DHS surveys back to 1994 #46

[athowes]

• ZWE PHIA has zero in sexcohab
• TZA PHIA has zero in sexcohab

[athowes]

Fixed those zeros in sexcohab by changing the sexbehav-vars-adam branch of naomi.utils. All good now I think!