X-Git-Url: https://pintos-os.org/cgi-bin/gitweb.cgi?a=blobdiff_plain;f=src%2Flanguage%2Fstats%2Freliability.c;h=90baa51fe2b861539291996bb6286858150b4232;hb=912833f9331784d692ade66c76bffb5497b890e0;hp=c73790713cebaceedf0bdc42a87e594eaa5cc43f;hpb=c0e45e85a654b657583164b3595994794138cd9e;p=pspp diff --git a/src/language/stats/reliability.c b/src/language/stats/reliability.c index c73790713c..90baa51fe2 100644 --- a/src/language/stats/reliability.c +++ b/src/language/stats/reliability.c @@ -1,5 +1,5 @@ /* PSPP - a program for statistical analysis. - Copyright (C) 2009, 2010 Free Software Foundation, Inc. + Copyright (C) 2009, 2010, 2011, 2013, 2015 Free Software Foundation, Inc. This program is free software: you can redistribute it and/or modify it under the terms of the GNU General Public License as published by @@ -18,26 +18,21 @@ #include -#include - -#include -#include - - -#include -#include -#include -#include -#include -#include - -#include -#include -#include - -#include -#include -#include +#include "data/casegrouper.h" +#include "data/casereader.h" +#include "data/dataset.h" +#include "data/dictionary.h" +#include "data/format.h" +#include "data/missing-values.h" +#include "language/command.h" +#include "language/lexer/lexer.h" +#include "language/lexer/variable-parser.h" +#include "libpspp/message.h" +#include "libpspp/misc.h" +#include "libpspp/str.h" +#include "math/moments.h" +#include "output/tab.h" +#include "output/text-item.h" #include "gettext.h" #define _(msgid) gettext (msgid) @@ -113,6 +108,27 @@ struct reliability static bool run_reliability (struct dataset *ds, const struct reliability *reliability); +static void +reliability_destroy (struct reliability *rel) +{ + int j; + ds_destroy (&rel->scale_name); + if (rel->sc) + for (j = 0; j < rel->n_sc ; ++j) + { + int x; + free (rel->sc[j].items); + moments1_destroy (rel->sc[j].total); + if (rel->sc[j].m) + for (x = 0; x < rel->sc[j].n_items; ++x) + free (rel->sc[j].m[x]); + free (rel->sc[j].m); + } + + free (rel->sc); + free (rel->variables); +} + int cmd_reliability (struct lexer *lexer, struct dataset *ds) { @@ -122,28 +138,30 @@ cmd_reliability (struct lexer *lexer, struct dataset *ds) reliability.n_variables = 0; reliability.variables = NULL; reliability.model = MODEL_ALPHA; - reliability.exclude = MV_ANY; + reliability.exclude = MV_ANY; reliability.summary = 0; - + reliability.n_sc = 0; + reliability.sc = NULL; reliability.wv = dict_get_weight (dict); - reliability.total_start = 0; + ds_init_empty (&reliability.scale_name); + - lex_match (lexer, '/'); + lex_match (lexer, T_SLASH); if (!lex_force_match_id (lexer, "VARIABLES")) { goto error; } - lex_match (lexer, '='); + lex_match (lexer, T_EQUALS); if (!parse_variables_const (lexer, dict, &reliability.variables, &reliability.n_variables, PV_NO_DUPLICATE | PV_NUMERIC)) goto error; if (reliability.n_variables < 2) - msg (MW, _("Reliabilty on a single variable is not useful.")); + msg (MW, _("Reliability on a single variable is not useful.")); { @@ -154,7 +172,7 @@ cmd_reliability (struct lexer *lexer, struct dataset *ds) reliability.n_sc = 1; reliability.sc = xzalloc (sizeof (struct cronbach) * reliability.n_sc); - ds_init_cstr (&reliability.scale_name, "ANY"); + ds_assign_cstr (&reliability.scale_name, "ANY"); c = &reliability.sc[0]; c->n_items = reliability.n_variables; @@ -166,31 +184,31 @@ cmd_reliability (struct lexer *lexer, struct dataset *ds) - while (lex_token (lexer) != '.') + while (lex_token (lexer) != T_ENDCMD) { - lex_match (lexer, '/'); + lex_match (lexer, T_SLASH); if (lex_match_id (lexer, "SCALE")) { struct const_var_set *vs; - if ( ! lex_force_match (lexer, '(')) + if ( ! lex_force_match (lexer, T_LPAREN)) goto error; if ( ! lex_force_string (lexer) ) goto error; - ds_init_string (&reliability.scale_name, lex_tokstr (lexer)); + ds_assign_substring (&reliability.scale_name, lex_tokss (lexer)); lex_get (lexer); - if ( ! lex_force_match (lexer, ')')) + if ( ! lex_force_match (lexer, T_RPAREN)) goto error; - lex_match (lexer, '='); + lex_match (lexer, T_EQUALS); vs = const_var_set_create_from_array (reliability.variables, reliability.n_variables); - + free (reliability.sc->items); if (!parse_const_var_set_vars (lexer, vs, &reliability.sc->items, &reliability.sc->n_items, 0)) { const_var_set_destroy (vs); @@ -201,7 +219,7 @@ cmd_reliability (struct lexer *lexer, struct dataset *ds) } else if (lex_match_id (lexer, "MODEL")) { - lex_match (lexer, '='); + lex_match (lexer, T_EQUALS); if (lex_match_id (lexer, "ALPHA")) { reliability.model = MODEL_ALPHA; @@ -211,12 +229,12 @@ cmd_reliability (struct lexer *lexer, struct dataset *ds) reliability.model = MODEL_SPLIT; reliability.split_point = -1; - if ( lex_match (lexer, '(')) + if ( lex_match (lexer, T_LPAREN) + && lex_force_num (lexer)) { - lex_force_num (lexer); reliability.split_point = lex_number (lexer); lex_get (lexer); - lex_force_match (lexer, ')'); + lex_force_match (lexer, T_RPAREN); } } else @@ -224,7 +242,7 @@ cmd_reliability (struct lexer *lexer, struct dataset *ds) } else if (lex_match_id (lexer, "SUMMARY")) { - lex_match (lexer, '='); + lex_match (lexer, T_EQUALS); if (lex_match_id (lexer, "TOTAL")) { reliability.summary |= SUMMARY_TOTAL; @@ -238,8 +256,8 @@ cmd_reliability (struct lexer *lexer, struct dataset *ds) } else if (lex_match_id (lexer, "MISSING")) { - lex_match (lexer, '='); - while (lex_token (lexer) != '.' && lex_token (lexer) != '/') + lex_match (lexer, T_EQUALS); + while (lex_token (lexer) != T_ENDCMD && lex_token (lexer) != T_SLASH) { if (lex_match_id (lexer, "INCLUDE")) { @@ -256,6 +274,14 @@ cmd_reliability (struct lexer *lexer, struct dataset *ds) } } } + else if (lex_match_id (lexer, "STATISTICS")) + { + lex_match (lexer, T_EQUALS); + msg (SW, _("The STATISTICS subcommand is not yet implemented. " + "No statistics will be produced.")); + while (lex_match (lexer, T_ID)) + continue; + } else { lex_error (lexer, NULL); @@ -268,6 +294,12 @@ cmd_reliability (struct lexer *lexer, struct dataset *ds) int i; const struct cronbach *s; + if ( reliability.split_point >= reliability.n_variables) + { + msg (ME, _("The split point must be less than the number of variables")); + goto error; + } + reliability.n_sc += 2 ; reliability.sc = xrealloc (reliability.sc, sizeof (struct cronbach) * reliability.n_sc); @@ -324,11 +356,11 @@ cmd_reliability (struct lexer *lexer, struct dataset *ds) if ( ! run_reliability (ds, &reliability)) goto error; - free (reliability.variables); + reliability_destroy (&reliability); return CMD_SUCCESS; error: - free (reliability.variables); + reliability_destroy (&reliability); return CMD_FAILURE; } @@ -351,6 +383,19 @@ run_reliability (struct dataset *ds, const struct reliability *reliability) struct casereader *group; struct casegrouper *grouper = casegrouper_create_splits (proc_open (ds), dict); + int si; + + for (si = 0 ; si < reliability->n_sc; ++si) + { + struct cronbach *s = &reliability->sc[si]; + int i; + + s->m = xzalloc (sizeof *s->m * s->n_items); + s->total = moments1_create (MOMENT_VARIANCE); + + for (i = 0 ; i < s->n_items ; ++i ) + s->m[i] = moments1_create (MOMENT_VARIANCE); + } while (casegrouper_get_next_group (grouper, &group)) @@ -415,11 +460,10 @@ do_reliability (struct casereader *input, struct dataset *ds, { struct cronbach *s = &rel->sc[si]; - s->m = xzalloc (sizeof (s->m) * s->n_items); - s->total = moments1_create (MOMENT_VARIANCE); + moments1_clear (s->total); for (i = 0 ; i < s->n_items ; ++i ) - s->m[i] = moments1_create (MOMENT_VARIANCE); + moments1_clear (s->m[i]); } input = casereader_create_filter_missing (input, @@ -477,7 +521,7 @@ do_reliability (struct casereader *input, struct dataset *ds, alpha (s->n_items, s->sum_of_variances, s->variance_of_sums); } - text_item_submit (text_item_create_format (TEXT_ITEM_PARAGRAPH, "Scale: %s", + text_item_submit (text_item_create_format (TEXT_ITEM_PARAGRAPH, _("Scale: %s"), ds_cstr (&rel->scale_name))); case_processing_summary (n_valid, n_missing, dataset_dict (ds)); @@ -501,6 +545,7 @@ case_processing_summary (casenumber n_valid, casenumber n_missing, int heading_rows = 1; struct tab_table *tbl; tbl = tab_create (n_cols, n_rows); + tab_set_format (tbl, RC_WEIGHT, wfmt); tab_headers (tbl, heading_columns, 0, heading_rows, 0); tab_title (tbl, _("Case Processing Summary")); @@ -545,27 +590,27 @@ case_processing_summary (casenumber n_valid, casenumber n_missing, total = n_missing + n_valid; tab_double (tbl, 2, heading_rows, TAB_RIGHT, - n_valid, wfmt); + n_valid, NULL, RC_WEIGHT); tab_double (tbl, 2, heading_rows + 1, TAB_RIGHT, - n_missing, wfmt); + n_missing, NULL, RC_WEIGHT); tab_double (tbl, 2, heading_rows + 2, TAB_RIGHT, - total, wfmt); + total, NULL, RC_WEIGHT); tab_double (tbl, 3, heading_rows, TAB_RIGHT, - 100 * n_valid / (double) total, NULL); + 100 * n_valid / (double) total, NULL, RC_OTHER); tab_double (tbl, 3, heading_rows + 1, TAB_RIGHT, - 100 * n_missing / (double) total, NULL); + 100 * n_missing / (double) total, NULL, RC_OTHER); tab_double (tbl, 3, heading_rows + 2, TAB_RIGHT, - 100 * total / (double) total, NULL); + 100 * total / (double) total, NULL, RC_OTHER); tab_submit (tbl); @@ -581,8 +626,10 @@ reliability_summary_total (const struct reliability *rel) const int heading_columns = 1; const int heading_rows = 1; const int n_rows = rel->sc[0].n_items + heading_rows ; - + const struct variable *wv = rel->wv; + const struct fmt_spec *wfmt = wv ? var_get_print_format (wv) : & F_8_0; struct tab_table *tbl = tab_create (n_cols, n_rows); + tab_set_format (tbl, RC_WEIGHT, wfmt); tab_headers (tbl, heading_columns, 0, heading_rows, 0); tab_title (tbl, _("Item-Total Statistics")); @@ -631,13 +678,13 @@ reliability_summary_total (const struct reliability *rel) moments1_calculate (s->total, &weight, &mean, &var, 0, 0); tab_double (tbl, 1, heading_rows + i, TAB_RIGHT, - mean, NULL); + mean, NULL, RC_OTHER); tab_double (tbl, 2, heading_rows + i, TAB_RIGHT, - s->variance_of_sums, NULL); + s->variance_of_sums, NULL, RC_OTHER); tab_double (tbl, 4, heading_rows + i, TAB_RIGHT, - s->alpha, NULL); + s->alpha, NULL, RC_OTHER); moments1_calculate (rel->sc[0].m[i], &weight, &mean, &var, 0,0); @@ -648,7 +695,7 @@ reliability_summary_total (const struct reliability *rel) tab_double (tbl, 3, heading_rows + i, TAB_RIGHT, - item_to_total_r, NULL); + item_to_total_r, NULL, RC_OTHER); } @@ -686,8 +733,11 @@ reliability_statistics (const struct reliability *rel) int n_rows = rol[rel->model].n_rows; int heading_columns = rol[rel->model].heading_cols; int heading_rows = rol[rel->model].heading_rows; - + const struct variable *wv = rel->wv; + const struct fmt_spec *wfmt = wv ? var_get_print_format (wv) : & F_8_0; struct tab_table *tbl = tab_create (n_cols, n_rows); + tab_set_format (tbl, RC_WEIGHT, wfmt); + tab_headers (tbl, heading_columns, 0, heading_rows, 0); tab_title (tbl, _("Reliability Statistics")); @@ -724,9 +774,6 @@ static void reliability_statistics_model_alpha (struct tab_table *tbl, const struct reliability *rel) { - const struct variable *wv = rel->wv; - const struct fmt_spec *wfmt = wv ? var_get_print_format (wv) : & F_8_0; - const struct cronbach *s = &rel->sc[0]; tab_text (tbl, 0, 0, TAB_CENTER | TAT_TITLE, @@ -735,9 +782,9 @@ reliability_statistics_model_alpha (struct tab_table *tbl, tab_text (tbl, 1, 0, TAB_CENTER | TAT_TITLE, _("N of Items")); - tab_double (tbl, 0, 1, TAB_RIGHT, s->alpha, NULL); + tab_double (tbl, 0, 1, TAB_RIGHT, s->alpha, NULL, RC_OTHER); - tab_double (tbl, 1, 1, TAB_RIGHT, s->n_items, wfmt); + tab_double (tbl, 1, 1, TAB_RIGHT, s->n_items, NULL, RC_WEIGHT); } @@ -745,9 +792,6 @@ static void reliability_statistics_model_split (struct tab_table *tbl, const struct reliability *rel) { - const struct variable *wv = rel->wv; - const struct fmt_spec *wfmt = wv ? var_get_print_format (wv) : & F_8_0; - tab_text (tbl, 0, 0, TAB_LEFT, _("Cronbach's Alpha")); @@ -760,8 +804,6 @@ reliability_statistics_model_split (struct tab_table *tbl, tab_text (tbl, 2, 1, TAB_LEFT, _("N of Items")); - - tab_text (tbl, 1, 2, TAB_LEFT, _("Part 2")); @@ -771,15 +813,12 @@ reliability_statistics_model_split (struct tab_table *tbl, tab_text (tbl, 2, 3, TAB_LEFT, _("N of Items")); - - tab_text (tbl, 1, 4, TAB_LEFT, _("Total N of Items")); tab_text (tbl, 0, 5, TAB_LEFT, _("Correlation Between Forms")); - tab_text (tbl, 0, 6, TAB_LEFT, _("Spearman-Brown Coefficient")); @@ -795,14 +834,14 @@ reliability_statistics_model_split (struct tab_table *tbl, - tab_double (tbl, 3, 0, TAB_RIGHT, rel->sc[1].alpha, NULL); - tab_double (tbl, 3, 2, TAB_RIGHT, rel->sc[2].alpha, NULL); + tab_double (tbl, 3, 0, TAB_RIGHT, rel->sc[1].alpha, NULL, RC_OTHER); + tab_double (tbl, 3, 2, TAB_RIGHT, rel->sc[2].alpha, NULL, RC_OTHER); - tab_double (tbl, 3, 1, TAB_RIGHT, rel->sc[1].n_items, wfmt); - tab_double (tbl, 3, 3, TAB_RIGHT, rel->sc[2].n_items, wfmt); + tab_double (tbl, 3, 1, TAB_RIGHT, rel->sc[1].n_items, NULL, RC_WEIGHT); + tab_double (tbl, 3, 3, TAB_RIGHT, rel->sc[2].n_items, NULL, RC_WEIGHT); tab_double (tbl, 3, 4, TAB_RIGHT, - rel->sc[1].n_items + rel->sc[2].n_items, wfmt); + rel->sc[1].n_items + rel->sc[2].n_items, NULL, RC_WEIGHT); { /* R is the correlation between the two parts */ @@ -821,12 +860,12 @@ reliability_statistics_model_split (struct tab_table *tbl, r /= sqrt (rel->sc[2].variance_of_sums); r /= 2.0; - tab_double (tbl, 3, 5, TAB_RIGHT, r, NULL); + tab_double (tbl, 3, 5, TAB_RIGHT, r, NULL, RC_OTHER); /* Equal length Spearman-Brown Coefficient */ - tab_double (tbl, 3, 6, TAB_RIGHT, 2 * r / (1.0 + r), NULL); + tab_double (tbl, 3, 6, TAB_RIGHT, 2 * r / (1.0 + r), NULL, RC_OTHER); - tab_double (tbl, 3, 8, TAB_RIGHT, g, NULL); + tab_double (tbl, 3, 8, TAB_RIGHT, g, NULL, RC_OTHER); tmp = (1.0 - r*r) * rel->sc[1].n_items * rel->sc[2].n_items / pow2 (rel->sc[0].n_items); @@ -835,7 +874,7 @@ reliability_statistics_model_split (struct tab_table *tbl, uly -= pow2 (r); uly /= 2 * tmp; - tab_double (tbl, 3, 7, TAB_RIGHT, uly, NULL); + tab_double (tbl, 3, 7, TAB_RIGHT, uly, NULL, RC_OTHER); } }